Grid: Difference between revisions

From DISI
Jump to navigation Jump to search
No edit summary
No edit summary
Line 1: Line 1:
    AUTHOR: Todd Ewing (based on work by Elaine Meng and Brian Shoichet)
AUTHOR: Todd Ewing (based on work by Elaine Meng and Brian Shoichet)


    USAGE: grid -i grid.in [-stv] [-o grid.out]
USAGE: grid -i grid.in [-stv] [-o grid.out]


    OPTIONS:
OPTIONS:
    - i input_file #Input parameters extracted from input_file, or grid.in if not specified
- i input_file #Input parameters extracted from input_file, or grid.in if not specified
    -o output_file #Output written to output_file, or grid.out if not specified
-o output_file #Output written to output_file, or grid.out if not specified
    -s Input parameters entered interactively
-s Input parameters entered interactively
    -t Reduced output level
-t Reduced output level
    -v Increased output level
-v Increased output level


    DESCRIPTION:
= Overview =
Grid creates the grid files necessary for rapid score evaluation in DOCK. Two types of scoring are available: contact and energy scoring. The scoring grids are stored in files ending in *. cnt and *. nrg respectively. When docking, each scoring function is applied independent of the others and the results are written to separate output files. Grid also computes a bump grid which identifies whether a ligand atom is in severe steric overlap with a receptor atom. The bump grid is identified with a *.bmp file extension. The file containing the bump grid also stores the size, position and grid spacing of all the grids.


    3.1.1. Overview
The grid calculation must be performed prior to docking. The calculation can take up to 45 minutes, but needs to be done only once for each receptor site. Since DOCK can perform continuum scoring without a grid, the grid calculation is not always required. However, for most docking tasks, such as when multiple binding modes for a molecule or multiple molecules are considered, it will become more time efficient to precompute the scoring grids.


    Grid creates the grid files necessary for rapid score evaluation in DOCK. Two types of scoring are available: contact and energy scoring. The scoring grids are stored in files ending in *. cnt and *. nrg respectively. When docking, each scoring function is applied independent of the others and the results are written to separate output files. Grid also computes a bump grid which identifies whether a ligand atom is in severe steric overlap with a receptor atom. The bump grid is identified with a *.bmp file extension. The file containing the bump grid also stores the size, position and grid spacing of all the grids.
NOTE: The following parameter definitions will use the format below:


    The grid calculation must be performed prior to docking. The calculation can take up to 45 minutes, but needs to be done only once for each receptor site. Since DOCK can perform continuum scoring without a grid, the grid calculation is not always required. However, for most docking tasks, such as when multiple binding modes for a molecule or multiple molecules are considered, it will become more time efficient to precompute the scoring grids.
parameter_name [default] (value):
#description


    NOTE: The following parameter definitions will use the format below:
In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.


        parameter_name [default] (value):
General Grid Parameters
        #description
 
    In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.
 
    General Grid Parameters


         * compute_grids [no] (yes, no):
         * compute_grids [no] (yes, no):
Line 45: Line 42:
           #Prefix for file name of grids. File extension will be appended automatically.
           #Prefix for file name of grids. File extension will be appended automatically.


    RETURN TO TABLE OF CONTENTS
= Bump Checking =
 
Prior to scoring, each orientation can be processed with the bump filter to reject ones that penetrate deep into the receptor. Orientations that pass the bump filter are then scored and/or minimized with any of the available scoring functions. A bump is based on the sum of the van der Waals radii of the two interacting atoms. The user specifies what fraction of the sum is considered a bump. For example, the default definition of a bump is if any two atoms approach closer than 0.75 of the sum of their radii. Grid stores an atomic radius which corresponds to smallest radius of ligand atom at the grid position which would still trigger a bump. During docking, for a given orientation, the position of each atom is checked with the bump grid. If the radius of the atom is greater than or equal to the radius stored in the bump grid, then the atom triggers a bump. To conserve disk space, the atom radius is multiplied by 10 and converted to a short unsigned integer.
    3.1.2. Bump Checking
 
    Prior to scoring, each orientation can be processed with the bump filter to reject ones that penetrate deep into the receptor. Orientations that pass the bump filter are then scored and/or minimized with any of the available scoring functions. A bump is based on the sum of the van der Waals radii of the two interacting atoms. The user specifies what fraction of the sum is considered a bump. For example, the default definition of a bump is if any two atoms approach closer than 0.75 of the sum of their radii. Grid stores an atomic radius which corresponds to smallest radius of ligand atom at the grid position which would still trigger a bump. During docking, for a given orientation, the position of each atom is checked with the bump grid. If the radius of the atom is greater than or equal to the radius stored in the bump grid, then the atom triggers a bump. To conserve disk space, the atom radius is multiplied by 10 and converted to a short unsigned integer.


    NOTE: The following parameter definitions will use the format below:
NOTE: The following parameter definitions will use the format below:


        parameter_name [default] (value):
parameter_name [default] (value):
        #description
#description


     In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.
     In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.
Line 72: Line 66:
[[Image:Contact_score.jpg|center]]
[[Image:Contact_score.jpg|center]]


    3.1.3. Contact Scoring
=Contact Scoring =
 
Contact scoring in grid incorporates the scoring performed with the DISTMAP program (developed by Shoichet and Bodian). The score is a summation of the heavy atom contacts (every atom except hydrogen) between the ligand and receptor. A contact is defined as an approach of two atoms within some cutoff distance (usually 4.5 Angstroms). If the two atoms approach close enough to bump (as identified with the bump grid) then the interaction can be penalized by an amount specified by the user.
    Contact scoring in grid incorporates the scoring performed with the DISTMAP program (developed by Shoichet and Bodian). The score is a summation of the heavy atom contacts (every atom except hydrogen) between the ligand and receptor. A contact is defined as an approach of two atoms within some cutoff distance (usually 4.5 Angstroms). If the two atoms approach close enough to bump (as identified with the bump grid) then the interaction can be penalized by an amount specified by the user.


Distance dependence of contact score function
Distance dependence of contact score function
Line 94: Line 87:
           #Maximum distance between heavy atoms for the interaction to be counted as a contact
           #Maximum distance between heavy atoms for the interaction to be counted as a contact


RETURN TO TABLE OF CONTENTS
= Energy scoring =
 
    3.1.4. Energy scoring


     The energy scoring component of DOCK is based on the implementation of force field scoring. Force field scores are approximate molecular mechanics interaction energies, consisting of van der Waals and electrostatic components:
     The energy scoring component of DOCK is based on the implementation of force field scoring. Force field scores are approximate molecular mechanics interaction energies, consisting of van der Waals and electrostatic components:
Line 121: Line 112:
[[Image:Grid_vdw_term.jpg|center]]
[[Image:Grid_vdw_term.jpg|center]]


Precomputing potentials on a grid
== Precomputing potentials on a grid ==
 
By inspection of the above equations , the repulsion and attraction parameters ( Aij and Bij ) for the interactions of identical atoms can be derived from the van der Waals radius, R, and the well depth, e.
    By inspection of the above equations , the repulsion and attraction parameters ( Aij and Bij ) for the interactions of identical atoms can be derived from the van der Waals radius, R, and the well depth, e.


    In order to evaluate the interaction energy quickly, the van der Waals and electrostatic potentials are precomputed for the receptor and stored on a grid of points containing the docking site. Precomputing the van der Waals potential requires the use of a geometric mean approximation for the A and B terms, as shown:
In order to evaluate the interaction energy quickly, the van der Waals and electrostatic potentials are precomputed for the receptor and stored on a grid of points containing the docking site. Precomputing the van der Waals potential requires the use of a geometric mean approximation for the A and B terms, as shown:


    Using this approximation, the first equation can be rewritten:
Using this approximation, the first equation can be rewritten:


    Three values are stored for every grid point k ,each a sum over receptor atoms that are within a user defined cutoff distance of the point:
Three values are stored for every grid point k ,each a sum over receptor atoms that are within a user defined cutoff distance of the point:


    These values, with trilinear interpolation, are multiplied by the appropriate ligand values to give the interaction energy. Grid calculates the grid values and stores them in files. The values are read in during a DOCK run and used for force field scoring.
These values, with trilinear interpolation, are multiplied by the appropriate ligand values to give the interaction energy. Grid calculates the grid values and stores them in files. The values are read in during a DOCK run and used for force field scoring.


    The user determines the location and dimensions of the grid box using the program showbox (see showbox). It is not necessary for the whole receptor to be enclosed; only the regions where ligand atoms may be placed need to be included. The box merely delimits the space where grid points are located, and does not cause receptor atoms to be excluded from the calculation. Besides a direct specification of coordinates, there is an option to center the grid at a sphere cluster center of mass. Any combination of spacing and x, y, and z extents may be used.
The user determines the location and dimensions of the grid box using the program showbox (see showbox). It is not necessary for the whole receptor to be enclosed; only the regions where ligand atoms may be placed need to be included. The box merely delimits the space where grid points are located, and does not cause receptor atoms to be excluded from the calculation. Besides a direct specification of coordinates, there is an option to center the grid at a sphere cluster center of mass. Any combination of spacing and x, y, and z extents may be used.


     NOTE: The following parameter definitions will use the format below:
     NOTE: The following parameter definitions will use the format below:

Revision as of 08:44, 30 November 2007

AUTHOR: Todd Ewing (based on work by Elaine Meng and Brian Shoichet)

USAGE: grid -i grid.in [-stv] [-o grid.out]

OPTIONS: - i input_file #Input parameters extracted from input_file, or grid.in if not specified -o output_file #Output written to output_file, or grid.out if not specified -s Input parameters entered interactively -t Reduced output level -v Increased output level

Overview

Grid creates the grid files necessary for rapid score evaluation in DOCK. Two types of scoring are available: contact and energy scoring. The scoring grids are stored in files ending in *. cnt and *. nrg respectively. When docking, each scoring function is applied independent of the others and the results are written to separate output files. Grid also computes a bump grid which identifies whether a ligand atom is in severe steric overlap with a receptor atom. The bump grid is identified with a *.bmp file extension. The file containing the bump grid also stores the size, position and grid spacing of all the grids.

The grid calculation must be performed prior to docking. The calculation can take up to 45 minutes, but needs to be done only once for each receptor site. Since DOCK can perform continuum scoring without a grid, the grid calculation is not always required. However, for most docking tasks, such as when multiple binding modes for a molecule or multiple molecules are considered, it will become more time efficient to precompute the scoring grids.

NOTE: The following parameter definitions will use the format below:

parameter_name [default] (value):

  1. description

In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.

General Grid Parameters

       * compute_grids [no] (yes, no):
         #Flag to compute scoring grids
             o grid_spacing [0.3] (float):
               # The distance between grid points along each axis.
       * output_molecule [no] (yes, no):
         #Flag to write out the coordinates of the receptor into a new, cleaned-up file. Atoms are
         #resorted to put all residue atoms together
             o receptor_out_file [receptor_out.mol2] (string):
               #File for cleaned-up receptor when output_molecule set
       * receptor_file [receptor.mol2] (string):
         #Receptor coordinate file. Partial charges and atom types need to be present.
       * box_file [site_box.pdb] (float):
         #File containing SHOWBOX output file which specifies boundaries of grid
       * vdw_definition_file [vdw.defn] (string):
         #VDW parameter file (see vdw.defn)
       * score_grid_prefix [grid] (string):
         #Prefix for file name of grids. File extension will be appended automatically.

Bump Checking

Prior to scoring, each orientation can be processed with the bump filter to reject ones that penetrate deep into the receptor. Orientations that pass the bump filter are then scored and/or minimized with any of the available scoring functions. A bump is based on the sum of the van der Waals radii of the two interacting atoms. The user specifies what fraction of the sum is considered a bump. For example, the default definition of a bump is if any two atoms approach closer than 0.75 of the sum of their radii. Grid stores an atomic radius which corresponds to smallest radius of ligand atom at the grid position which would still trigger a bump. During docking, for a given orientation, the position of each atom is checked with the bump grid. If the radius of the atom is greater than or equal to the radius stored in the bump grid, then the atom triggers a bump. To conserve disk space, the atom radius is multiplied by 10 and converted to a short unsigned integer.

NOTE: The following parameter definitions will use the format below:

parameter_name [default] (value):

  1. description
   In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.
   Bump Grid Parameters
       * bump_filter [no] (yes, no):
         #Flag to screen each orientation for clashes with receptor prior to scoring and minimizing
       * bump_overlap [0.75] (float):
         #Amount of VDW overlap allowed. If the probe atom and the receptor heavy atom approach
         #closer than this fraction of the sum of their VDW radii, then the position is flagged as a
         #bump
                 0 = Complete overlap allowed
                 1 = No overlap allowed
Contact score.jpg

Contact Scoring

Contact scoring in grid incorporates the scoring performed with the DISTMAP program (developed by Shoichet and Bodian). The score is a summation of the heavy atom contacts (every atom except hydrogen) between the ligand and receptor. A contact is defined as an approach of two atoms within some cutoff distance (usually 4.5 Angstroms). If the two atoms approach close enough to bump (as identified with the bump grid) then the interaction can be penalized by an amount specified by the user.

Distance dependence of contact score function

   The attractive score in grid is negative and a repulsive score is positive. This switch of sign is necessary to allow the same minimization protocol to be used for contact scoring as implemented for energy scoring.
   NOTE: The following parameter definitions will use the format below:
       parameter_name [default] (value):
       #description
   In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.
   Contact Grid Parameters
       * contact_score [no] (yes, no):
         #Flag to construct contact grid
       * contact_cutoff_distance [4.5] (float):
         #Maximum distance between heavy atoms for the interaction to be counted as a contact

Energy scoring

   The energy scoring component of DOCK is based on the implementation of force field scoring. Force field scores are approximate molecular mechanics interaction energies, consisting of van der Waals and electrostatic components:
   where each term is a double sum over ligand atoms i and receptor atoms j, which include the quantities listed below.

Generalization of the VDW component

   The van der Waals component of the scoring function has been generalized to handle any combination of repulsive and attractive exponents (providing that a> b). The user may choose to "soften" the potential by using a 6-9 Lennard -Jones function. The general form of the van der Waals interaction between two identical atoms is presented:
   where e is the well depth of the interaction energy, R is the van der Waals radius of the atoms, and coefficients C and D can be determined given the two following boundary conditions:

at

at

   Application of these boundary conditions to the above equation yields an expression of the van der Waals interaction with a generalized Lennard -Jones potential.
   The consequence of using a different exponent for the repulsive term is illustrated in Figure 1 . Notice that the well position and depth are unchanged, but that the repulsive barrier has shrunk by about a 0.25 Angstrom.

Distance dependence of the Lennard -Jones Function


Grid vdw term.jpg

Precomputing potentials on a grid

By inspection of the above equations , the repulsion and attraction parameters ( Aij and Bij ) for the interactions of identical atoms can be derived from the van der Waals radius, R, and the well depth, e.

In order to evaluate the interaction energy quickly, the van der Waals and electrostatic potentials are precomputed for the receptor and stored on a grid of points containing the docking site. Precomputing the van der Waals potential requires the use of a geometric mean approximation for the A and B terms, as shown:

Using this approximation, the first equation can be rewritten:

Three values are stored for every grid point k ,each a sum over receptor atoms that are within a user defined cutoff distance of the point:

These values, with trilinear interpolation, are multiplied by the appropriate ligand values to give the interaction energy. Grid calculates the grid values and stores them in files. The values are read in during a DOCK run and used for force field scoring.

The user determines the location and dimensions of the grid box using the program showbox (see showbox). It is not necessary for the whole receptor to be enclosed; only the regions where ligand atoms may be placed need to be included. The box merely delimits the space where grid points are located, and does not cause receptor atoms to be excluded from the calculation. Besides a direct specification of coordinates, there is an option to center the grid at a sphere cluster center of mass. Any combination of spacing and x, y, and z extents may be used.

   NOTE: The following parameter definitions will use the format below:
       parameter_name [default] (value):
       #description
   In some cases, parameters are only needed (questions will only be asked) if the parameter above is enforced. These parameters are indicated below by additional indentation.
   Energy Grid Parameters
       * energy_score [no] (yes, no):
         #Flag to perform energy scoring
       * energy_cutoff_distance [10] (float):
         #Maximum distance between two atoms for their contribution to the energy score to be
         #computed
       * atom_model [u] (u, a):
         # Flag for how to model of nonpolar hydrogens
           u = United atom model. Hydrogens attached to carbons are assigned a zero VDW well-depth and the partial charge is transferred to the carbon.
           a = All atom model. Hydrogens attached to carbons have regular VDW well-depth and partial charge is not modified.
       * attractive_exponent [6] (int):
         # Exponent of attractive Lennard - Jones term for VDW potential
       * repulsive_exponent [12] (int):
         # Repulsive of attractive Lennard - Jones term for VDW potential
       * distance_dielectric [yes] (yes, no):
         # Flag to make the dielectric depend linearly on the distance
       * dielectric_factor[4.0] (float):
         #Coefficient of the dielectric